Reactive Writers: How Co-Writing with AI Changes How We Engage with Ideas

A team of researchers from Cornell University—Advait Bhat, Marianne Aubin Le Quéré, Mor Naaman, and Maurice Jakesch—published a groundbreaking study titled 'Reactive Writers: How Co-Writing with AI Changes How We Engage with Ideas.' The paper, accepted to the CHI 2026 conference, presents a mixed-methods analysis combining retrospective interviews with 19 participants and a quantitative trace analysis of 1,291 distinct AI co-writing sessions. The core finding is a fundamental shift in the writing process: engaging with the AI's suggestions—reading and deciding whether to accept them—becomes the primary activity, taking time and cognitive focus away from traditional processes of personal ideation and language generation.

This shift leads to a new practice the researchers term 'Reactive Writing,' an evaluation-first, suggestion-led approach. Because writers often engage with AI suggestions before completing their own ideation, the AI's proposed ideas and opinions seed the direction of the text, which the writer then elaborates on. Crucially, the study found a significant disconnect: writers did not notice the AI's influence and reported feeling in full control, largely because they retained the technical ability to edit the final output. This combination makes 'Reactive Writing' highly vulnerable to subtle, AI-induced biases and opinion shifts, as the tool's suggestions can frame and steer the writer's thinking from the outset without their conscious awareness.